Prediction of Missing Values via Voting Ensemble

Data quality issue refers to the presence of an intolerable defect in dataset. Completeness, consistency, accuracy, validity, and timeliness are the five components that affects the data quality. Properly executing these components means achieving high quality and vice verse. Completeness of data is usually translated as non-existence of any missing values in the data. Missing data is one of the issues that directly affect the quality of the data and consequently the quality of decisions being made based on this data. The aim of this paper is to convert the problematic dataset into a complete one via a prediction framework to predict values for the missing data. Prediction of missing value via voting ensemble (PMVVE) framework is proposed to predict missing values via the utilization of multiple machine learning techniques to predict the missing values in time series data. Experimental results show that PMVVE have a better performance than other standard techniques that are usually utilized for filling the missing values.